Are we starting to cool: the case for the last three months

[Stratos Ferric]

In answer to a question posed by WiB, whose new found zeal for a potential post-modern cold dawn seems to know no bounds (though were I a betting man I suspect it will be firmly bounded by the next warm month we get), the following is an analysis of the current sequence of cold compared with the only two other similar warm spikes in the CET record.

The hypothesis is that the current warming is anomalous but natural, and that the very recent cooling might be the start of a trend. The conclusion, too early to say either way - but watch this space.

The chart contains a complex wad of data. I have taken three separate series as indicated, and plotted them linearly, showing, from the start of the series, the cumulative anomaly ©, the cumulative number of net warm months c.f. the rolling ten year mean for each month, and the actual monthly anomaly © for each month against the contemporary ten year rolling mean for that month. I've highlighted the main cooling sequences that followed each of the two warm spikes chosen, and also shown other "false dawns".

ERRATA

Can't be bothered to repost the plot: the previous signals is mislabelled and should read "...of cooling" not "...of warming".

Edited September 30, 2007 by Stratos Ferric

[memories of 63]

Nice chart SF. A quick question, in the 2007 chart, how can you put in data for September and October already?

[Stratos Ferric]

The data runs to the end of September and assumes a 13.9C outturn. October isn't in there. The final September figure may well vary slightly up or down from this level, but that variability is not material to the assessment.

[memories of 63]

The data runs to the end of September and assumes a 13.9C outturn. October isn't in there. The final September figure may well vary slightly up or down from this level, but that variability is not material to the assessment.

Thanks for the swift reply SF. I may be missreading your tables but in the chart in the bottom right at the bottom (2007) you have -1.0 for what I assume is September and 0.9 for what I can only interpret as October. Where am I going wrong?

[Devonian]

Thanks for the swift reply SF. I may be missreading your tables but in the chart in the bottom right at the bottom (2007) you have -1.0 for what I assume is September and 0.9 for what I can only interpret as October. Where am I going wrong?

I'm not sure I understand the data either.

It looks to me as if the data is shifted right one month - it was July 2006 that was +2.8 not August? But, some of the figures look odd as well.

SF, I don't disagree at all with you argument, but is there more to clarify here?

[Hiya]

I'm not sure I understand the data either.

It looks to me as if the data is shifted right one month - it was July 2006 that was +2.8 not August? But, some of the figures look odd as well.

SF, I don't disagree at all with you argument, but is there more to clarify here?

Yeah, same can be said for Feb/Mar 2006 they seem to be one month in advance.

Good plot tho.

Edited October 1, 2007 by Hiya

[Stratos Ferric]

Thanks for the swift reply SF. I may be missreading your tables but in the chart in the bottom right at the bottom (2007) you have -1.0 for what I assume is September and 0.9 for what I can only interpret as October. Where am I going wrong?

No, the chart runs horizontally Dec-Nov.

Yeah, same can be said for Feb/Mar 2006 they seem to be one month in advance.

Good plot tho.

Ahh, sorry, no, one thing I should have made clear but didn't...you've made an assumption which is a completely reasionable assumption to have made, but an assumption nonetheless. The data tables are configured Dec-Nov (not Jan-Dec): it makes it easier to tally the seasons in Excel without faffing between different rows, hence the appearance of a slip.

...

SF, I don't disagree at all with you argument, but is there more to clarify here?

I can't read your mind, ask away...

Edited October 1, 2007 by Stratos Ferric

[Devonian]

I can't read your mind, ask away...

You have clarified. Just my poor English.

[memories of 63]

No, the chart runs horizontally Dec-Nov.

Ahh, sorry, no, one thing I should have made clear but didn't...you've made an assumption which is a completely reasionable assumption to have made, but an assumption nonetheless. The data tables are configured Dec-Nov (not Jan-Dec): it makes it easier to tally the seasons in Excel without faffing between different rows, hence the appearance of a slip.

I can't read your mind, ask away...

SF thanks for the clarification - one query arises, however. You state that the September figure is estimated i.e. -0.9C this is not highlighted as a cool anomaly however if the posted figure is 0.1C lower then this will become part of the highlighted series. I don't know if this will affect the analysis but visually, it makes it appear more significant.

[Devonian]

SF thanks for the clarification - one query arises, however. You state that the September figure is estimated i.e. -0.9C this is not highlighted as a cool anomaly however if the posted figure is 0.1C lower then this will become part of the highlighted series. I don't know if this will affect the analysis but visually, it makes it appear more significant.

Now I'm getting confused again...

I thought September was just above average

Edit: hang on, ten year average?...

Edited October 1, 2007 by Devonian

[Stratos Ferric]

Now I'm getting confused again...

I thought September was just above average

Edit: hang on, ten year average?...

Yes, TEN year average. To be clear, the reason for using the shorter measure is that it reacts faster to changes. Yes, statistically there are fewer degrees of freedom, so drawing inferences requires us to use much wider margins, but the measure is more sensitive and will give an earlier indication of any change in trend. With the thirty year mean, such has been the warming across the thirty years, we could still have a big annual drop for any given month, yet still be higher than was the case 30 years ago: thus, the thirty year mean would still be increasing. The shorter time series corrects for much of this effect, but is also more volatile, because each data point has three times the gearing.

SF thanks for the clarification - one query arises, however. You state that the September figure is estimated i.e. -0.9C this is not highlighted as a cool anomaly however if the posted figure is 0.1C lower then this will become part of the highlighted series. I don't know if this will affect the analysis but visually, it makes it appear more significant.

It will make a difference in terms of the visual shading, for sure, but on the plot, which is the PROPER frame of reference, it makes bog all difference. This is precisely why, in determing any trends, we need to see a run of data emerge. It will be 2-3 months downstream before we can really start to draw any inferences about the three months past. What we can say for sure is that the rampant warming from Sept - May has been, at the very least, stopped for now. Whether it's a pause for breath or the begininnings of a correction down we can't yet tell. Statistically, the ten year mean is too far ahead of the thirty year mean now, so we're overdue a correction. The first 'nect interesting' point would be if the ten year mean were to fall below the thirty year mean, but that's a long way off.

[memories of 63]

Yes, TEN year average. To be clear, the reason for using the shorter measure is that it reacts faster to changes. Yes, statistically there are fewer degrees of freedom, so drawing inferences requires us to use much wider margins, but the measure is more sensitive and will give an earlier indication of any change in trend. With the thirty year mean, such has been the warming across the thirty years, we could still have a big annual drop for any given month, yet still be higher than was the case 30 years ago: thus, the thirty year mean would still be increasing. The shorter time series corrects for much of this effect, but is also more volatile, because each data point has three times the gearing.

It will make a difference in terms of the visual shading, for sure, but on the plot, which is the PROPER frame of reference, it makes bog all difference. This is precisely why, in determing any trends, we need to see a run of data emerge. It will be 2-3 months downstream before we can really start to draw any inferences about the three months past. What we can say for sure is that the rampant warming from Sept - May has been, at the very least, stopped for now. Whether it's a pause for breath or the begininnings of a correction down we can't yet tell. Statistically, the ten year mean is too far ahead of the thirty year mean now, so we're overdue a correction. The first 'nect interesting' point would be if the ten year mean were to fall below the thirty year mean, but that's a long way off.

Sf I gree with everything you've said. What would be even more interesting is to (arbitarily) choose a period where 'abnormal' warming started to affect the 30 year trend, isolate that time and see when the any future data start to raise or lower that figure. Unfortunately, if we're currently experiencing a 'long' natural cycle, I'll probably be dead by the time we see any change to the longer term averages

[Guest Viking141]

Hi SF

Very interesting stuff, thanks for that. As you say we would need to see several more cool months before there might be a cooling case to answer so I take it you are going to continue to plot this and keep us updated? That would be very interesting to see.

Edited October 1, 2007 by Viking141

[Hiya]

Yeah, same can be said for Feb/Mar 2006 they seem to be one month in advance.

Good plot tho.

Yeah the thought crossed my mind just afterwards, thanks anyway.

[Stratos Ferric]

Sf I gree with everything you've said. What would be even more interesting is to (arbitarily) choose a period where 'abnormal' warming started to affect the 30 year trend, isolate that time and see when the any future data start to raise or lower that figure. Unfortunately, if we're currently experiencing a 'long' natural cycle, I'll probably be dead by the time we see any change to the longer term averages

I think that, in fact, that is precisely what I've done. The periods being compared are the only ones in CET history where the ten year mean has risen sharply above the thirty year mean, albeit not from such a high starting point.

CET_summer_07_test.ppt

Hi SF

Very interesting stuff, thanks for that. As you say we would need to see several more cool months before there might be a cooling case to answer so I take it you are going to continue to plot this and keep us updated? That would be very interesting to see.

Certainly the plan: the advantage of having the analysis partition is that long period analyses don't disappear.

[Anti-Mild]

Impressive graph SF, and very interesting. As you say, time will tell.

Out of interest and off topic; what do you do for a living? Are you a statistician?

[West is Best]

Stratos - I hope this doesn't sound patronising but you don't know how heartening it is to find someone who generally believes in the warm-up at least fairly and reasonably to look at the evidence. I knew you had that in you. What I find difficult but intriguing are those who point-blank refuse even to look at some of the slightly out-of-kilter facts.

As you say, my own recent ? over the warm-up may well abate. But at least I'm trying to wrestle with what I am increasingly seeing as inconvenient facts to the universal GW argument. They may just be blips - as you say, time will tell.

Something I'd love to do, but don't have time is to slice up the cold period from autumn 2005 to April 2006 and paste that together with this year's cool off and then do the same for the warm year in the middle of those two and set the two side by side: one record warmth, the other at least pretty cold by modern standards.

Edited October 1, 2007 by West is Best

[Stratos Ferric]

Stratos - I hope this doesn't sound patronising but you don't know how heartening it is to find someone who generally believes in the warm-up at least fairly and reasonably to look at the evidence.

...

Something I'd love to do, but don't have time is to slice up the cold period from autumn 2005 to April 2006 and paste that together with this year's cool off and then do the same for the warm year in the middle of those two and set the two side by side: one record warmth, the other at least pretty cold by modern standards.

WiB: I only believe the warm up because that's what the evidence tells me. If the data says any different then as soon as I see it believe me, I'll say so.

Not quite sure what you'd be seeking to prove by the cut and paste: you're in danger of veering into Jethro territory there by wanting to omit inconvenient data. On a cursory basis the comparison you suggest actually stacks up well with the warming trend. The warm periods are particularly warm; the cool ones are less markedly cold.

What I may produce is a rolling series using something like 6, 12, 24, 60 and 120 month anomalies. That would be the equivalent of the charts used by chartists in the city to predict market movements. It would also provide a more granular hair trigger for assessing short term movements, and whether they are becoming medium term trends.

Impressive graph SF, and very interesting. As you say, time will tell.

Out of interest and off topic; what do you do for a living? Are you a statistician?

I could tell you but I'd have to shoot you. I give advice to the great and the good, but the job often entails analysis - though thankfully, nowadays, I tend to commission others to ride the spreadsheets for me.

[jethro]

WiB: I only believe the warm up because that's what the evidence tells me. If the data says any different then as soon as I see it believe me, I'll say so.

Not quite sure what you'd be seeking to prove by the cut and paste: you're in danger of veering into Jethro territory there by wanting to omit inconvenient data. On a cursory basis the comparison you suggest actually stacks up well with the warming trend. The warm periods are particularly warm; the cool ones are less markedly cold.

What I may produce is a rolling series using something like 6, 12, 24, 60 and 120 month anomalies. That would be the equivalent of the charts used by chartists in the city to predict market movements. It would also provide a more granular hair trigger for assessing short term movements, and whether they are becoming medium term trends.

I could tell you but I'd have to shoot you. I give advice to the great and the good, but the job often entails analysis - though thankfully, nowadays, I tend to commission others to ride the spreadsheets for me.

Whoa, hang on a second there Stratos! The only times I have ever said "let's remove data" it has been, as I have explained at the time and again afterwards, has been to demonstrate a point and from an academic perspective - not dismiss as though it had never happened. From memory this has been about two points; the cooler interlude in the 60's 70's which I backed with scientific evidence which supports my questions about global dimming affecting the CET and global temps, it did although when I posted my backing evidence you failed to respond further. Ditto when talking about global temps and the high point of 0.6 degree increase in 1998 - an anomoly which I backed up with supporting evidence again from reputable sources which state catagorically the anomoly was caused by the immense El Nino that year. Again, you did not respond further. I do not omit inconvenient data, I raise questions with supporting evidence. To suggest otherwise, as you have done above is dismissive, disingenuous and incorrect.

[Stratos Ferric]

Whoa, hang on a second there Stratos! The only times I have ever said "let's remove data" it has been, as I have explained at the time and again afterwards, has been to demonstrate a point and from an academic perspective - not dismiss as though it had never happened. From memory this has been about two points; the cooler interlude in the 60's 70's which I backed with scientific evidence which supports my questions about global dimming affecting the CET and global temps, it did although when I posted my backing evidence you failed to respond further. Ditto when talking about global temps and the high point of 0.6 degree increase in 1998 - an anomoly which I backed up with supporting evidence again from reputable sources which state catagorically the anomoly was caused by the immense El Nino that year. Again, you did not respond further. I do not omit inconvenient data, I raise questions with supporting evidence. To suggest otherwise, as you have done above is dismissive, disingenuous and incorrect.

Sorry if that rubbed you up the wrong way Jethro, but as I mentioned in the thread the other night, it's a dangerous road we go down if we start ascribing to outlying data characteristics of any sort that allow us to dismiss the data, no matter HOW objective the assessment might seem. You're forever asking whether or not we're really warming (as opposed to going through a heightened blip), and how we can know how much, if any, of the current warmth is driven by GHG emissions. Applying the same logic, why should we be certain about the precise degree of warming caused by a nino?

Leave the data in and provide a narrative, that way the audience can make their own mind up. Removing data for any reason, however objective we are, or think we are being, is always going to look like mischief making to a trained observer.

[jethro]

Sorry if that rubbed you up the wrong way Jethro, but as I mentioned in the thread the other night, it's a dangerous road we go down if we start ascribing to outlying data characteristics of any sort that allow us to dismiss the data, no matter HOW objective the assessment might seem. You're forever asking whether or not we're really warming (as opposed to going through a heightened blip), and how we can know how much, if any, of the current warmth is driven by GHG emissions. Applying the same logic, why should we be certain about the precise degree of warming caused by a nino?

Leave the data in and provide a narrative, that way the audience can make their own mind up. Removing data for any reason, however objective we are, or think we are being, is always going to look like mischief making to a trained observer.

Yes it did rub me up the wrong way as it was suggestive of mischief making. I can see the point you make but we approach this from different angles, you say leave it in to judge the overall warming, I say can we isolate a proportion which we know can be ascribed to causes other than Co2. I actually, as I have said time and time again have NEVER asked are we warming, is this a blip? I have only ever asked how much of this warming is down to Co2? We approach from different angles but we seek the same answer surely? If you and others can post links, use information from accredited sources to support their stance, then why can't I? If I don't then I'm wrong - nowt to back it up with and yet if I do, you then say how can we be certain about the precise degree of warming caused by a Nino. Because if we are to accept the findings of scientists to support the theory of AGW then we must also accept the reports of scientists who measure the ENSO cycle. If you are to throw one accredited source out on that point, then we must throw them all out. Debate with me the facts and science Stratos but please do not malign me as you did earlier.

[Roger J Smith]

If you want to say that May to September was basically a near-normal period, without much prolific warmth at any point, that tends to indicate that almost any current theory is holding its own -- long-term warming held in check, warming fading out, dramatic changes afoot, etc etc. Only time will tell which is valid. That's my read on it, anyway, I don't think it indicates anything very clear-cut in any direction. It is not of course a dramatic signal of runaway global warming, so that needs to be duly noted as I think it has been by none other than West is Best. But whatever lies around the corner, this spell of rather cool by recent standards but essentially near-normal temperature, is not a smoking gun of any kind, in my (very) humble opinion.

If it means anything, I could add that since about November 2006, we have been stuck in a very similar rut of near-normal, seemingly cool weather (by recent standards) on this west-facing coast as well. So that's a bit longer than your spell, and through it all, there have been occasional windstorms and deluges, but no really cold weather by standards of the past 50 or 100 years either. Just a lot of drab, nondescript weather that comes in very close to the 30-year normals.

It just isn't a dramatic turn of any kind, and I would have to say it feels like a lull in the upward climb to me, but winter temperatures are of course more variable than other seasons, and so if this is a precursor to colder weather, then something dramatic could lie ahead. I think the chances are about as high as they have been in recent years for this to take place, in either your climate region or mine. It has also been a long time since a very cold winter here, we did manage a prolific snowstorm last winter (on 29-30 November) that lasted about a week, and the same thing happened between Christmas and New Years in 1996. Otherwise, it has just been the usual one week of cold weather that we seem to get every winter, nothing like say 1950 when the whole of January was well below freezing here. I think it was 1990 that last had even a ghost of that sort of winter freeze on this west coast.

[JOHN.MAC]

as i posted in the winter discussion i was read an article on a website the other day and it was predicting that the NAO would be below average for this winter would this trend towards a colder than average winter?

[Guest Viking141]

Whilst we would obviously need to see more evidence before coming to any sort of conclusions, it does look, up here at least, that the warming may have at least flatlined. September was another below average month but it was also the coolest since 1986 in Shetland. It occurs to me that I may see the evidence of a "cooldown" first given my extreme northerly location.

The temps (compared against the long-term mean)for the last five months have been as follows:

May : Average

June : Average

July : Below Average

August: Below Average

September: Below Average

Edited October 22, 2007 by Viking141

[Altohumorous]

How's it looking so far for October? My impression has been of quite a warm month overall, only a few chilly nights, no real frosts, and the prospect of a continuing very mild trend over the next week (taking night-time values into account)